Refrigerator Circuit Anomaly

I have been monitoring my refrigerator circuit for eight months and just noticed a strange anomaly from four months ago. Typically, the compressor comes on several times a day, drawing 90-100 watts. The ice maker uses a 210 watt resistance heater for 2-3 minutes before dumping ice. However, on August 4 at 11:30am, a steady ~1500 watt load appeared and remained for about 14 hours. The refrigerator compressor and ice maker appeared to continue functioning normally during that time.

At first I assumed it was a measurement error of some sort because the refrigerator is on a dedicated circuit with nothing else plugged in to that outlet behind the refrigerator. However, a second CT on the mains also recorded the same 1500 watt load. If it was actually a real load, I have no idea what it could have been.

Background and additional information:
My home was built in 1991 in Richmond, Virginia, USA. I believe the electrical code at the time would have required the refrigerator to be on a dedicated circuit. I used a tone generator to confirm that no other outlet inside or outside the house is fed from that circuit. My utility supplies 240v split-phase power. When I set up my IoTaWatt and emonCMS, I called one phase “A” and the other phase “B.” I have a model ECS24200 CTs on each phase of the mains coming into my load center. I then have several model ECS1050 CTs on individual circuits. The refrigerator is a 120v circuit on phase B and the mysterious load showed up on both the refrigerator CT as well as the phase B mains CT. My IoTaWatt is currently on firmware version 02_04_02 and has always been set to MINOR auto-update class.

The average outdoor temperature that day was 80°F (27°C) with a high of 91°F (33°C) and a low of 73°F (23°C). Nobody was home when the mystery load started at 11:30am and the security system was armed so it is highly unlikely that anyone entered the home during the day. My partner returned home around 9-10pm that evening but did not notice anything unusual (no smell of smoke, no thawed food in the freezer, etc.). In fact, I only noticed this anomaly last week when I was looking back at my recorded data.

My refrigerator is a Whirlpool model WRT318FZDM02, manufactured in December 2017. Its full load amp rating is 6.5 @ 115V, which only accounts for 748 watts. The profile of the mystery load appears to be that of an electric resistance heater. However, there are only a few devices in the home that are designed to pull 1500 watts or more.

  1. A 10kW backup electric resistance heater in our heat pump air handler is on a dedicated circuit, monitored via a dedicated CT, and did not turn on during that time.

  2. The electric oven/range is on a dedicated circuit, monitored via a dedicated CT, and did not turn on during that time.

  3. Our electric heat pump water heater is on a dedicated circuit, monitored via two dedicated CTs (IoTaWatt and built-in CT), and is set to “heat pump only” mode so it can only pull about 450 watts.

  4. We have a washing machine and heat pump dryer that share a dedicated 240v circuit that I installed myself and are monitored by a dedicated CT.

  5. We own one portable 1500-watt space heater which was put away in storage at the time.

Several appliances can pull up to 1100 watts (dishwasher, toaster oven, and pressure cooker) but I can’t think of any more devices that can sustain 1500 watts for 14 hours with no inrush current. I would appreciate any help in solving this mystery. Let me know if I can provide any more helpful information.

Here are four screenshots from emonCMS:

This one shows the one shows entire 8 months of historical data on the refrigerator circuit:

This one shows the phase B mains, refrigerator, dishwasher, and heat pump (indoor and outdoor units combined) during the time period in question:

This one shows just the refrigerator circuit during that time:

The mystery load started abruptly with no inrush current but ended in a strange intermittent way. This shows the intermittency at the end of the mystery load:

I wouldn’t dispute that. The “turn-off” looks as if it could be a very dodgy thermostat opening.

The extra energy is about 20 kWh, that’s several bathtubs of hot water, so if that went in heat, I reckon you’d know about it. But does the heat pump show a commensurate change in consumption? Could it be that the fridge door was left open just enough to have it run continuously and yet be able to maintain temperature? If that, why didn’t it stop at 9:30 pm when somebody could have come in and seen it? I know you’ve said that it’s rated at only half that, so it’s a long shot, but one to consider.

Realistically, that’s a deliberate load, not a wiring fault per se. But it might be wires going somewhere else.

Being really cynical, does “no other outlet inside or outside the house is fed from that circuit” mean exactly that, or could it mean “no other outlet inside or outside the house that I know of is fed from that circuit”, i.e. nobody has sneaked in a connection to something somewhere that you don’t know about - not that day, but at some time in the past 30 years?

Did your neighbours notice anything unusual in the area at the same time?

The simple fact that two c.t’s and two inputs saw it seems to rule out a straightforward internal fault - hardware or software - in your kit.

I’ve been studying that last graph for a while and I’ve just noticed - it doesn’t add up. Either the separate inputs are drawn to different scales, or your inputs aren’t calibrated, because the heights of the individual lines (yellow, red & green) add up to more than the blue. It’s especially noticeable between 9 & 10 am, when the total is less than the heat pump on its own. You’ve not mentioned PV generation - is that the explanation?

First, I’d like to thank Ben for the time he took to provide a nicely detailed bit of “extra info.”
(If other users would follow suit, it woud definitely help us to help them. hint, hint)

Given he’s (as near as he knows and his tone gen check) got nothing else on that circuit
and the CT attached to (I’m going to guess) an SEW, it looks as if the culprit is indeed the “fridge.”

It sounds like the compressor was close to being in a locked rotor state, if not actually locked.
Whether or not that would have caused a circuit breaker to trip, or thermal cutout to activate, I can’t say. In the linked video (kind of cheesy, but gets the point across) the guys cut open a bad compressor
to reveal a “locked” pump, that is easily “broken free” to spin again, only to lock up again.
Long shot, I know, but sometimes truth is stranger than fiction.

The refrigerator I have has similar specs WRT to current draw. My T.E.D. 5000 has a fast enough
response that it spikes to ~1200 Watts during compressor start-up. Normal running load is about 205-210 Watts as measured with a Kill-A-Watt energy monitor. Hence my guess about a compressor that’s “dying a slow death.”

The air conditioner units that cool the radar I maintain act this way when their compressors are on the
verge of failure. One of the first things the HVAC guys do is measure the motor current with a clamp-on
ammeter. In 7 years, I’ve seen it happen at least three times. That’s not counting the instances where
one of the other members of our crew (five in total) have seen the same thing. We have two sites
with a total of 8 aircon units, and it can get “fairly warm” in Oklahoma during the summer. :wink:

I think I’ve got to say that something like 15× the normal running current, sustained for 14 hours, means that there’s something desperately wrong with the protection on that fridge. Fortunately, for the sake of the contents :open_mouth: If that’s really what might have happened - and it sounds as plausible as anything - I hope there’s nothing combustible near that compressor.

But it cured itself and has run for another 4+ months?

That was the weird part of the video. They could spin the compressor crankshaft by hand and
it would do a few revs then stop. When it stopped, it required a bit of effort to get it to spin again,
but not a great deal, as they were able to “break it free” again, by hand.
The spinning/stopping/spinning was “repeatable,” which made it all the more oddball.

There is one other possibility - a second resistance heater within the appliance? If it is marketed as a frost free product, the way the automatic defrosting is achieved is with a resistance element within the walls of the fridge or freezer, with a timer / thermostat / PCB control & thermistor to terminate the defrost. The defrosted water then trickles down the back onto a tray sat above the compressor at the back of the machine to then evaporate away with the heat of the compressor.

I am pretty sure that there have been malfunctioning defrost stats / defrost timers on some fridges which resulted in them going up in smoke. If yours was running for 14 hours though, it would soon melt the fridge, so a locked rotor compressor could be more likely.

It’s Christmas - treat yourself to a more efficient fridge! This one is clearly burning a hole in your wallet. Lets hope it doesn’t do any more than that.

I think you’ve hit it, MJ.

Even though it ran for 14 hours, the OP said the compressor cycled normally. That would have kept
the inside temp near normal, yet explain the resistive load as well as the flaky “shutoff.” (bad stat)

Hmm, weirder yet…

The defrost heater, having a typical power rating of 350 W to 600 W,
Ref: Auto-defrost - Wikipedia

Typical power rating… Maybe he’s got one that has a defrost heater “on steroids.” :thinking:

Resistance heater + compressor running together?

The graph shows changes of 200-300 watts during the period in question. Looks quite plausible!

I’d go with that too. Seeing the cycling behind the “heater” load is what worried me about the locked-rotor theory. And the second heater also explains why the protection didn’t work - the load was expected (by the manufacturer if not by us) so it shouldn’t have worked.

Exactly, that is a lot of energy and, if it is real, then it ended up as heat somewhere. If it ended up inside the refrigerator then I think we would have lost all our food because there is no way that 100 watt compressor would have been able to remove the 1500 watts of heat as it was being delivered.

You are correct… there is no other outlet that I know of.

You are right again… I have a CT on each of the Mains (A & B) but I am only showing phase B because that’s the phase where the mystery load showed up. Since the heat pump is a balanced 240v load, I capture that with a single CT and double the value. If you take half of the Heat Pump Total, then it starts to add up. Let me know if there is a more straightforward way to configure my IoTaWatt to avoid confusion like that.

I wondered this myself but then I saw the power profile of the compressor sitting on top of the 1500 watt load. That leads me to believe that the compressor was behaving normally before, during, and after the 14-hour period.

An electrical fault of some sort is my biggest concern. If the energy is going somewhere useful, that’s fine. If not, then there is potential for a fire and I want to know how and where.

I think you are on to something. The one thing I keep coming back to is where did all that heat go? If it went inside the fridge, wouldn’t the compressor have had to run non-stop for a much longer time to remove that much heat?

Here is a screenshot showing the full week surrounding the event:

It just seems like the compressor cycling is relatively unaffected by the mystery load.

I think the chances that it’s not real are slim. If it isn’t real, explain how it got onto the main c.t and no other circuit? If it affected one circuit or all, then I’d say it is feasible that it’s a fault in the hardware or software, but not as it affected two with a physical link between them.

Is there any evidence of the heat going into the house, is what I meant when referring to the heat pump - assuming it was cooling given the outside ambient? That still fits with a second heater - just that the heat was going outside the fridge rather than inside. That then raises the question: for what purpose? To evaporate condensation run-off? 20 kWh? ! ! ! No.

Ah, I did wonder that. But given the nature of the anomaly, it was worth checking.

There are very few of us here who know anything about that. Lemaire has his own forum.

From the owners manual, which I think is for the correct product:

owners-manual.pdf (2.8 MB)


It recommends a “15 or 20 A fused” supply. Wouldn’t 15A have been insufficient?

The plot thickens…

15A breaker might not be sufficient if the fridge was designed to draw over 1500 watts. However, it would be sufficient if it only has a “normal” defrost heater.

I downloaded the wiring diagram which describes the two types of defrost cycle controllers. Mine behaves like the Electronic Timer defrost with ~400 watt defrosts showing up every day or two.
WRT318FZD Wiring Diagram.pdf (702.8 KB)

I understand now. And after looking further into my data, it doesn’t appear that the heat was even generated inside the house.

The week before this incident, I happened to measure the real-world efficiency of my AC with some new enthalpy probes I got for work. I measured my AC at a COP of 2.7 with an outdoor temperature of 93°F. That means my AC would have to use an additional 7.4 kWh of energy to remove the 20 kWh of heat added by the mystery load (20 kWh / 2.7). However, that extra energy does not show up on my AC system that day. My heat pump used 18.5 kWh for the whole day which is actually less than the summer average when normalized for outdoor temperature.

I took daily heat pump energy use for all of July and August and divided it by daily Cooling Degree Days (CDD - base 65°F). For those two months, my worst day was 2.63 kWh/CDD, my best day was 1.08 kWh/CDD, and my average was 1.47 kWh/CDD. August 4th was 1.22 kWh/CDD which is below average. Keep in mind that no one was home most of the day so internal gains were marginally less and the thermostat was set at 75°F rather than 74°F, but these are small differences. If I subtract the 7.4 kWh, that I would expect to see, from the 18.5 kWh used that day, I’m left with 11.1 kWh. When I divide that by the 15.2 CDD, I get 0.73 kWh/CDD, which is unrealistically lower than any other day that summer.

This leads me to believe that the mystery load must have been outside the house. It also makes me question again if it is real or not.

I agree that the chances of measurement hardware or software fault are slim but the chances of a real load that large for that long are becoming increasingly slim as well! What gives?

Do you feel it’s worth heading over to The IoTaWatt forum and asking about the likelihood of a hardware or software fault there? I think I know what the response will be - it’s your call.

Am I correct to assume that all the c.t’s are in your load centre and the Iotawartt is also close by? How far apart are the two c.t’s in question (the B leg feeder and the fridge circuit)?

There’s nothing on the A leg that corresponds with the mystery load?

Coming back to my wild theory about mystery wires: Do you know the actual route the cable takes from your load centre to the socket? and is it possible it’s been “interfered” with? I’m now trying to think of what sort of signal could be injected into that circuit, that would look like a 1.5 kW load to the iotawatt and not appear on any other c.t.

just a thought for you as to mysterious load-- do any of your wires exist in damp or wet location… I had a mysterious load issue a few years back. turns out it was trickle charge of 300w across a 120 volt wire in a damp location. when the area dried it disappeared and when it got damp it came back.( strung a new underground wire and it stopped) … you live in the UK so your wires voltage ~230V so you can carry a significant higher trickle load then what I would at 120V which might account for the unaccountable usage … and that long duration was just the length of time it took for that area to dry from heat dissipation of the trickle charge to stop – curious does UK electrical code not require individual fuse/breaker just for a the fridge

Stephen, it’s a 120 V US circuit. Benjamin has his Country of Residence as “United States”, and in the top post he specifically states it’s on the “B” leg of a 240 V split-phase supply, which I believe is your standard.

So in that case, you’re looking at a leakage current of around 11.6 A, representing a shunt resistance of about 10 Ω. It was a good thought, but I don’t think the numbers stack up.

In the UK, for domestic premises and offices, it’s normal to have a “final ring sub-circuit” where an unlimited number of socket outlets are connected in a ring (serving not more than 100 m² of floor area), the ring starting and ending at the “consumer unit”, where it’s protected by a 32 A MCB. Each plug contains a fuse rated at not more than 13 A. It’s usual for the whole house to be protected by a 30 mA RCD, so more than 30 mA leakage to earth (but not Line - Neutral) would result in an RCD trip.

So, excepting cookers and the like, every UK domestic appliance will be fed in that way.

oops my mistake I must of assumed a it was UK-- or I pick it up from another poster and transposed the location…
where I live the standard is outlet is every 8 - 10 feet for 100 feet of length to a max of 120 feet wire length - and no more then 3 light sockets per fuse if you have maximum number of outlets on that string – it probably increased now with the advent of energy efficient lights and the banning of sale of incandescent lights . fridges, microwaves have to have dedicated outlet. and kitchen outlet no more the 4 halves of an outlet ( using 14/3 wire and 2 15 amp breakers

:laughing: That is my favourite trick.

We use 2.5 mm² cable, rated at around 23 A (depending on how it’s installed) but of course each socket is fed from both sides of the loop, so a socket roughly in the middle will have two feeds in parallel, each capable of 23 A, and the current will share roughly equally.

I was wondering if I could do the maths to see just how much water 20 kWh of energy would raise from (say) 28 °C and turn into steam. From here it’s about 0.3 MJ/kg to raise it to 100 °C, and then 2.5 MJ/kg to evaporate it. So 0.77 kWh per kg. I reckon that our 20 kWh of energy that’s gone missing would boil away to nothing 25 kg (litres) of water - neglecting losses, so in real life somewhat less.

When I built my house, the Leccy and I had a serious conversation about doing star layout using 4mm² cable for the sockets. In some places that was what we ended up doing.